DOCUMENT RESUME

ED 130 211 CG 010 892

AUTHOR Zimmerman, Barry J.; Rosenthal, Ted L.TITLE Concept Attainment, Transfer, and Retention Through

Observation and Rule Provision.INSTITUTION Arizona Univ., Tucson. Arizona Center for Early

Childhood Education.PUB DATE Nov 71VOTE 23p.; Not available in hard copy due to marginal

legibility of original document

EDRS PRICE MF-$0.83 Plus Postage. MC Not Available from EDRS.DESCRIPTORS Cognitive Processes; *Concept Formation; Concept

Teaching; Elementary Education; *Elementary SchoolStudents; Instructional Systems; *Learning Processes;*Observational Learning; *Retention Studies; RoleModels; Stimulus Generalization; *Transfer ofTraining

ABSTRACTThe effects of observing a model and of providing a

response rule on the learning, transfer, and retention of adial-reading, numerical concept were studied in 144 third-graders.Different experimenters conducted the immediate learning proceduresversus the measurements of retention. No extrinsic reinforcers werepromised or dispensed. The children profited both from modeling andfrom rule-provision, with the strongest learning, transfer, andretention displayed by the group that watched the model and alsoreceived the rule summary. Sequence of presenting the sets ofretention stimuli (including a series of novel generalization itemsnot previously encountered) dia not influence the strength of conceptretention six weeks after training. (Author)

***********************************************************************Documents acquired by ERIC include many informal unpublished

* materials not available from other sources. ERIC makes every effort ** to obtain the best copy available. Nevertheless, items of marginal ** reproducibility are often encountered and this affects the quality ** of the microfiche and hardcopy reproductions-ERIC makes available ** via the ERIC Document Reproduction Service (EDRS). EDRS is not* responsible for the quality of the original document. Reproductions ** supplied by EDRS are the best that can be made from the original.***********************************************************************

V,,,Ert,-,==.=:Tn?..Y:[17 T.::,E.VV.2-Xf."5.,Zre....

1Jt Iill

.V.

1

0vt 4

11

1,1

V; 1

fi0

34,

il

Pi

g1,1

1'A

'''=164=XXIStatVC.....SalZCZ---1=`^'-=17'..71=7.1,215,171,Z1NISP.M5.41.7,1".;EIST.I. "

ARIZONA CENTERFOR EARLY CHILDHOOD EDUCATIONCollege of EducationUniversity of Arizona1515 East First StreetTucson, Arizona 85719

CONCEPT ATTAINMENT, TRANSFER, ANDRETENTION THROUGH OBSERVATION ANDRULE-PROVISION

By: Barry J. Zimmerman andTed. L. Rosenthal

November, 1971

Zimmerman 1

Abstract

The effects of observing a model and of providing a response rule on

the-leran-ing,/itransfer, and retention of a dial-reading, numerical concept

were studied in 144 third-graders. Different experimenters conducted the

immediate learning procedures versus the measurements of retention. No

extrinsic reinforcers were promised or dispensed. The children profited

both from modeling and from rule-provision, with the strongest learning,

transfer, and retention displayed by the group that watched the model and7

also received the rule summary. Sequence of presenting the sets of retention

stimuli (including a series of novel generalization items not previously

encountered) did not influence the strength of concept retention six weeks

after training.

CONCEPT ATTAINMENT, TRANSFEA, AND RETENTION THROUGH

OBSERVATION AND RULE-PROVISION1

Barry 3. Zimmerman and Ted. L. Rosenthal

University of Arizona

Current research on social learning phenomena has been increasingly

directed to the cognitive realm. One main thrust has involved the modfi-

cation of children's language responses, and here modeling has been shown

to affect the formation of simple sentences (Carrol, Rosenthal, & Brysh,

in press; Rosenthal & Whitebook, 1970), more complex sentence rubrics

(Harris & Hassemer, in press; Rosenthal & Carroll, in press), the production

of rule-consistent types of questions (Rosenthal & Zimmerman, in press a;

Rosenthal, Zimmerman, & Durning, 1970), and the use of selected construc-

tions, such as prepositional phrases, both natural (Bandura & Harris, 1966;

Odom, Liebert, & Hill, 1968).and ungrammatical (Liebert, Odom, Hill, & Huff,

1969) in form. In certain of these experiments, the children not only

"imitated- the model's usages but, without any additional training, trans-.

ferred the rule-consistent pattern to new stimuli.

A second main concern has dealt with concept-formation. The' observa-

tional learning and transfer of conceptual responses has been demonstrated

with multidimensional conservation tasks (Rosenthal & Zimmerman, in press

b), with a simple equivalence problem (Rosenthal, Moore, Dorfman, & Nelson

1971), and with a much more demanding stimulus discrimination problem

Zimmerman 3

(Zimmerman & Rosenthal, in press). All these studies assessed short-term

effects. A necessary next step is to examine relationships between obser-

vational learning and later memory. One must determine if vicariously-in-

stated responses maintain stability well after training has been completed.

In a previous experiment, Rosenthal, Alford, and Rasp (in press), us-

ing a complex clustering task, found that modeling created substantial im-

mediate concept acquisition and transfer. After several weeks' delay, the

children were retested and retention proved to depend on the combination

of modeling and simultaneous verbal coding which they had witnessed earlier.

Although evidence of retention was obtained, concept maintenance required

both observation of the model's cluster and stronger mnemonic verbal cod-

ing during her demonstrations. The children who only observed the model per-

form silently did not significantly exceed their baseline mean when reassessed.

It thus seemed important to confirm that, by itself, observation of a model

can produce retention of an abstract paradigm.

In the previous design, the effect of providing a rule to guide response

was not wholly separated from verbal coding as the model executed her re-

sponse demonstrations. Although children who both observed a strong verbal

code and later received a summary rule showed the best immediate learning

and transfer, their performance declined after delay, relative to the strong

code, no rule group. However, control children given the rule with no other

training did show some later concept retention. Therefore, it seemed desir-__

able to study rule-provision, with and without modeling, when the additional

influence of mnemonic codes was eliminated. It was predicted that modeling

and rule-provision would each assist learning, with the strongest effects

occurring when the two procedures were combined.

Zimmerman 4

Several other features of the present operations merit attention as

steps to expand the applicability of conceptual social learning. The prior

study used two stimulus sets to assess both immediate learning and retention.

If a concept (e.g. telling time) has been learned and retained with some

stability, a child should be able to utilize his knowledge when new dial--

surrounds are enCountered, as with clock faces that differ in color, shape,

and spatial proportions of dial, hands, and numerals. For this reason,

along with the sets of stimuli used in immediate learning, the present re-

tention tests included a new set of stimuli whose concept7irrelevant visual

details were very different from those previously presented. Indeed, the

social learning viewpoint assumz z-hat conceptual paradigms are abstracted

which, through judicious selection and inference, then guide response to

specific stimulus instances (e.g. Rosenthal & White, in press). In contrast,

a more mechanical view of "association" ,dould, presumably, expect that prior

pairing should establish decidedly stronger response strengths to the in-

itial training stimuli, and to the order of stimulus presentation'that most

closely approximated the initial conditions of learning (see Helton, 1963).

Thus, the sequence of administering the three sets of retention stimuli was

systematically varied, and the effect of differences in order was analyzed.

Unlike the previous clustering concept (which emphasized the object

class of stimuli), the present concept required the child to coordinate

the color and dial-position of an arrow to the color and number of response

items, thus giving more prominence to a variable, numerical attribute.

Finally, all the retention daa were collected by an experimenter who

did not know the design or hypotheses of the study, who was unaware of the

children's scores during original learning, and who was a stranger to the

children. The retention data were thus freed from any putative expectancy

7

Zimmerman5

effects on the part of the experimenter (e.g. Rosenthal, 1969), from depen-

dence on the personalities of particular adults, and from the sorts of sub-

tle context effects discussed by Postman (1968), in which presumably-negli-

gible constancies, like the same experimenter, can sometimes become attached

as discriminative stimuli that partly maintain response strength.

METHOD

Subjects and Experimenters

From third grade classes at three schools serving middle-income Anglo-

American areas of Tucson, 72 boys and 72 girls were randomly drawn and as-

signed to each factorial combination of treatments. The children ranged in

age from 7.5 to 10.5 years, with a mean age of 8.6 years. A female graduate

student served as experimenter and a male graduate student served as the

model during immediate learning procedures; a different female experimenter

collected the delayed data. All adults were Anglo-Americans in-their twen-

ties, with no striking departures from average characteristics.

Task Naterials

Three sets of stimulus cardboards we-re prepared. The first set, which

was used in baseline, for training, and for the immediate and delayed imi

tation phases, comprised 2 X 2 in. square outlines drawn in black india ink

on white cards. From the border of the square base a perpendicular, one in.

arrow was drawn in one of four positions: up, down, left, or right. These

arrows occurred in three colors, blue, red, or yellow (BRY), and the combi-

nation of colors X positions thus created 12 BRY-squares stimulus cards.

The response materials were 18 sewing spools, six being painted in each of

the BRY colors. The rule required coordinating the arrow-position to a

number, -gith up = 1, left = 2, down = 3 and right = 4, analogous to a dial

Zimmerman6

with four counter-clockwise positions to specify the number of spools to be

used. The arrow's color denoted the color for spool selection. Thus, with

a red arrow in the up position, one red spool was the correct response; with

a blue arrow in the left position, two blue spools was the correct response;

and with a yellow arrow in the right position, four yellow spools was the

correct response, etc. The second set of stimulus cards, (BRY-circles),

was used for immediate and delayed transfer. These stimuli had circular bases

of one in. radius from which projected arrows 1/2 in. long in the same four

positions, and blue, red and yellow colors as above. The response materials

(BRY spools), and the iu12 relating arrow-position and color to correct

response were identical.

The third set of stimuli was used only for a delayed test of stringent

generalization, in which the colors of both stimulus and response materials,

as well as the basal shape of the stimuli, were varied. The stimulus bases

were "diamonds" (rhomboids) of one in. radius from which, in the same four

positions, projected two in. arrows in the colors green, orange, and purple

(GOP). The rule for correctly coordinating arrow color and position to

response was the same, but the response spools for the GOP stimuli were painted

in the green, orange, and purple colors.

Thus, although the basic relation of dial position and color to spool-

selection remained constant, the task materials varied in the basal shapes

of the stimuli, in the length of the arrow (equal to the radius of BRY-squares;

one-half the radius of BRY-circles; twice the radius of GOP-diamonds), and

in the artow and spool colors between the BRY and GOP materials. All stimuli

were drawn on 8 X 8 in. white cards which were mounted in ring-binders.

Procedure and Design

Immediate learning and transfer. The child was taken individually from

class to a test room by the experimenter, introduced to the model, and was

directed by the experimenter as follows: "We're going to play a game of

picking spools. First, I'm going to show you a set of picture cards, one

at a time. For each card, you try and guess how many spools are the right

answer. Sometimes the right answer-is just one spool, but most of the time

the right answer needs more than one spool. Here's the first card; now you

pick one or more spools, and put,your spool-guess here (on a sheet of paper).

Now put your spools back in the box and I'll show you the next card, etc."

After baseline, the no-model, no-rule control group was instructed as

follows: "Now that you have had a chance to get to know the game better,

we'll give you another turn with the same,pictures. Here's the first card;

put your spool-guess on this paper, etc." The BRY-squarcs were then read-

ministered. These same directions were also given to the no-model, with

rule-provision group which also received the following rule information be-

fore responding to the stimuli: "Now, before your next turn, let me give

you a good rule for playing this game. The arrow moves like a backward clock

and tells you how many spools to pick. When the arrow is up, you need just

one spool; when the arrow is left, you need two spools; when the arrow is

down you need three spools; when the arrow is right you need four spools.

The color of the arrow tells you which color of spools to pick to be correct."

Modeling groups were instructed after baseline as follows: "Okay, this

man is going to show you a good way to play this game. You watch him care-

fully and you'll have another turn later." Themodel then selected the cor-

rect numbers and colors of spools for each card while the child observed.

1 0

Zimmerman8

The model performed silently throughbut and received neither praise nor feed-

back from the experimenter. Subsequent to the model's demonstration, the

BRY-squares task was again given to the model, no-rule group. The experi-

menter provided the same rule directions (as above) to the model plus rule

group, before readministering the BM-squares task.

Next, with no further training, the BRY-circles were introduced, And

all children were instructed as follows: "Now we'll play the game vith

some new pictures. For each card, you try to guess or figure out how many

spools are the right suer. Here's the first card; pick the number of

spools you .think are right and then put them on this paper, etc." Upon

completing this first transfer task, the child was thanked, praised for his

performance, and returned to class. Thus, for immediate learning, the design

involved a 2 (rule, no-rule) X 2 (modeling, no-modeling) X 3 (baseline, imi-

tation, and transfer trials) factorial with 18 boys and 18 girls in each

cell.

Delayed retention and generalization. After some six weeks, the second

(new) experimenter returned to the school and reassessed the children in

roughly the same order in which they were initially studied. The new ex-

perimenter was naive with respect to the nature of the rule (and simply

recorded each child's actual spool choices per card), to the prior experimental

treatment given the child, and to his previous performance. The sequence of

presenting the BM-squares, the BRY-circles, and the new, GOP-diamonds was

systematically varied so that, by random assignment, 3 boys and 3 girls

from.each condition of original training were exposed to each of the sixA

Ossible.,orders'of pretenting the three sets of delayed phase stimuli.,

Whatevei. the 'order,of ititaltispresentation, the experimenter intro-

duced the first set of stimuli as follows: NRay, yot probably remember

ii-

Zimmerman 9

that a few weeks ago you played a game of picking spools. Today we're going

to play that same game again; probably some of the pictures you see today

will be familiar to you; other pictures will be new. Anyway, you remember

that I will show you a set of picture cards, one at a time. For each card,

you try to remember, or figure out, how many spools are the right answer.

Here's the first card; now you pick one or more spools from this box (point-

ing) and put your spools here (on a sheet of paper), etc."

Both the BRY and GOP spools were visible to the child, but in separate

boxes, and the experimenter pointed to the spools of the same color as the

stimuli in giving her instructions. After the first delayed trials, the

second and the final sets of stimuli were each introduced by the experimenter

as follows: "Here are some other cards; keep on picking spools from this

box (pointing), etc."

The design compared the originfA baseline scores to performance on the

delayed tasks, across presentatiof3 ,3equence, for the modeling and rule varia-

tions in a 2 (rule, no/rule) X 2 (modeling, no/modeling) X 6 (sequence) X

4 (trials) factorial design. For both immediate and delayed data, all post-

hoc comparisons (e.g. comparing baseline performance to the scores on the

BRY-squares, BRY-circles, or GOP diamonds, independent of their order of

presentation) were made with Tukey HSD tests (Kirk, 1963). Orthogonal compari-

sons by a multiple t-test (Kirk, 1958) were used to assess the performance

of the model, no-rule versus the rule, normodel.groups, and to compare the

average of these groups with the performance of the strongest, model plus

rule group, on all trials after baseline.

I 2

Zimmerman 10

RESULTS-

Immediate Data

For all phases, the means for each treatment group, and for subjects

combined on the basis of modeling and rule variations, are presented in

Table 1.

Insert Table 1 about here

From the overall analysis of variance; there was found a significant

trials effect showing increased concept-attainment from baseline to imita-

tion and to transfer phases (F = 118.70; df = 2/280; p < .001). Significant

main effects for modeling (F = 19.03; df = 1/140; p < .001) in favor of

children who had observed the model perform, and for rule-provision (F =

37.30 df = 1/140; p < .001) in favor of children who had received the rule,

also were obtained.

A significant modeling X phases interaction term (F = 12,92; df = 2/230;

p .< .001) was found. Analysis of this interaction by Tukey tests revealed

that the treatment conditions had not differed in baseline, and each varia-

tion increased its scores from baseline to imitation and to transfer phases

(all ps < .01), with no significant decline from imitation to transfer.

However, the modeling subjects outperformed non-modeling subjects in'the

imitation and the transfer phases (both ps < .01).

Similarly, rule provision interacted with phases (F = 33.52; df = 2/280;

l < .001). The rule and non-rule variations did not differ during baseline,

and each condition increased its scores significantly from baseline to im-

itation (both ps < .01). The rule-provided subjects also incrased from

1 3

Zimmerman. 11

baseline to transfer (p < .01), with no significant decline from tmitation to

transfer phases; in contrast, non-rule subjects did not significantly exceed

their baseline mean during transfer, and declined significantly (2_ < .05)

from imitation to the transfer phase. Further, the rule-provided subjects

surpassed the scores of their non-rule counterparts in the imitation and

transfer phases (both ps < .01). The three-way interaction term failed to

approach significance.

Orthogonal comparisons disclosed no difference between the rule, no-

model and the model, no-rule treatments during tmitation. However the model

plus rule group significantly surpassed (o < .01) the average of the two

foregoing conditions. In the transfer phase, the rule, no-model subjects out-

performed the model, no-rule subjects (p.< .01), but the model plus rule group

again significantly surpassed (p < .01) the average of the two other group.

Delayed Data

The phase means for the several treatment combinations are also pre-

.Sented in Table 1. The overall analysis of variance revealed a significant

trials effect (F =.47.94; df = 3/359; p < .001), showing that scores had in-

creased"from baseline to the delayed phases. Sir;nificant main .effects were

again obtained for modeling (F = 10.14; df = 1/120; y < .002) with children

who nad.observed the model scoring.higher, and for rule-provision (F = 16.20;

df = 1/120; p < .001) in favor of children who received the rule statement.

The sequence of presenting the stimulus sets (IRY-squares, IIRY-circles, and

GOP-diamonds) failed to create a main effect (F < 1.0, MS) or significant

interactions (largest F = 1.89; df = 5/120; p = .10), suggesting that the

children's informatiorrprocessing capacity was robust enough to overcome

differences in the order of presenting the delayed stimuli.

A si,,lnificant modeling X phases interaction term (F 4.33; df = 3/359;

Zimmerman12

p < .006) was found. Both the modeling and the non-modeling subjects (who

had not differed at baseline), displayed significant increases from base-

line to each of the three delayed phases (all ps < .01). Further Tukey

tests revealed that the modeling subjects outperformed the non-modeling

subjects in each of the delayed phases considered seperately (all Rs < .01).

Similarly, a sturkiicant_intex.action was found between rule-provision

and phases (F = 11.06; df = 3/359; p < .001). Both the rule-provided and

no-rule subjects (who had not-differed at baseline) increased significantly

from baseline to delayed imitation and to stringent generalization phases

(all ps < .01). However, whereas the rule-provided children also surpassed

their baseline mean in delayed transfer (p < .01), the no-rule subjects

failed to significantly exceed their baselines in the delayed transfer

phase. Further Tukey tests also supported the superiority of the rule-

provided children, who surpassed the no-rule condition in each of the three

delayed phaaes (all ps < .01). Excluding the sequence of presentation data,

already discussed, no other interaction terms approached significance (all

Fs < 1.0,

Orthogonal comparisons were applied to retention results. To dif-

ferences between the model, no-rule, no-model groups were found in any of

the delayed phases. However, the model plus rule group outperformed the

alierage of both outer groups with the imitation (p < .01), the transfer

(p < .01), and the stringent generalization (p < .05) stimuli.

DISCUSSION

To our knowledge, this is the first demonstration that from observation

alone, unaided by other means of conveying information, a concept can be

retained and generalized to novel stimuli after a substantial elapse of

time. Similarly, providing a correct verbal rule was an even briefer, and

Zimmerman 13

hence more efticient, method for producing concept learning, transfer, and

retention in third-graders under the present conditions. Empirically, model-

ing and rule-provision were not redundant operations the children who

received both procedures performed best in every phase; no interactions

involving the alternative operations jointly, nor their,combination with

any other variate, were found. It cannot be arguad that either training

method merely elicited readily available concept discriuinations because,

in such a case, modeling, rule-provision, or their sum should have created

response strengths near asymptote, in contrast to the actual findings.

The results further suggested that children acquired and stored a

mnemonic paradigm or cognitive rubric, rather than discrete links to par-

ticular physical stimuli. Thus, under short-term conditions, neither the

modeling nor the rule-provision subjects declined significantly from imi-

tation to transfer phases. Perhaps more striking, weeks later concept

retention was unaffected by the sequence of presenting the original train-__

ing, transfer, or the new--generalization--stimuli. If response were

mediated by narrowlydefined S - R "links", then the order of presen-

tation ,losest to prior experience should have given the strongest

retention, and presenting the novel stimuli first should have created

some discernible interference. Put otherwise, in any system emphasizing

stimulus features, generalization should drop off as the visible prop-

erties of test stimuli depart from those of training stimuli. In contrast,

if organized relationships are abstracted and remembered, then learning

should be relatively stable across a range of stimulus details if the

rule-governed motif remains invariant, as in the present data. Certainly,

without resort to inferential processes, one has difficulty explaining

1 6

Zimmerman 14

why the children who, after delay, first encountered the stringent gener-

alization items were able to take them in stride. Based on color, shape,

ratio of arrow to base, and before the more familiar items could "prop"

recall of the concept, one might have expected markedly reduced scores

with the new stimuli.

The foregoing argument is important for .;:ocial learning theory be-

cause of Bandura's (1969, in preas) views on acquisition through obser-.

vation. He assumes a mediated contiguity position in which, without

overt practice, stimulus-response chains are stored through covert, sym-

bolic processes. When the model displays a high standard for self-rein-

forcement, or a vigorous attack on a hapless Bobo doll, what is crucial is

the observer's reproduction of the behavior sequences demonstrated. How-

ever, when the model exemplifies rule-consistent, conceptual behavior, a

paradox of sorts arises. Literal emulation ("imitation') by itself might

result from copying of the motor_and.speech resaonses. modeled. It be-_

comes necessary to confirm, through tests of transfer, tnat symbolic par-.

adirJri has been learned. Having shown this, as in the present results and

the experiments cited earlier, one seeks to explain it in the same medi-

ated contiguity terms that Bandura has invoked for social learning gener-

ally. To do so, one must suppose that, at least in part, the child can

abstract the governing idea of the concept series and recognize its rele-

vance not only to the training stimuli, but to tile transfer stimuli as

well. P.ecourse to inference, Caen, was first required to maintain Bandura's

tneoretical viewpoint in the face of data showing conceptual transfer from

observational learning. It is both interesting and encouraging that re-

search partially aimed at defending a theoretical necessity has produced-

7

Zimmerman15

early evidence to favor an inferentially-mediated contiguity notion. Thus,

the absence of a relationship between presentation-sequence and Tetention

in the present study, results showing an increase in children's ability to

give a Correct response rule from observing a silent modeling procedure

(Rosenthal & Zimmerman,flin press b), and a variety of data questioning a

mechanical conception of word association (Rosenthal & Hhite, in press),

all shared as original impetus the goal of maintaining in tact basic prem-

ises of social learning theorizing by expanding its Conception of symbolic,

covert mediation to admit inferential reasoning. It is also reassuring

that the retention, and further generalization, of the dial-reading, nu-

merical concept was obtained by a wholly-naive experimenter, who could

hardly have exerted social influence or expectancy effects upon children

she had never met before.

1 8

Zimmerman 16

REFERENCES

Bandura, A. Principles of behavior modification. New York. Holt,

Rinehart & Winston, 1969.

Bandura, A. Social learning theory. New York: McCaleb-Seiler, in press

(1972).

Bandura, A., &.Harris, ii. B. liodification of syntatic style. Journal

of Experimental Child Psychology, 1966, A, 341-352.

Carroll, W. R,, Rosenthal, T. L., & Brysh, C. G. The social transmis-

sion of grammatical parameters. Journal of Educational Psychology,

63, in press (1972).

Harris, ,i. B., & Hassemer, t7. G. Some factors affecting the complexity

of children's sentences: the effects of modeling, age, sex, and

bilingualism. Journal of Experimental Child Psychology, in press

Experimental design: procedures for the behavioral sciences.

Belmont, Calif.: Brooks/Cole, 196S.

Liebert, R. M., Odom, R. D, Hill, J. H., & Huff, P. L. Effects of age

and rule familiarity on the production of modeled language construc-

tions. Developmental Psychology, ]Z)69, 1_, 108-112.

ilelton, A. N. Implications of short-term memory for a general theory of

memory. Journal of Verbal Learning and Verbal Behavior, 1963, 2, 1-21.

Odom, R. D., Liebert, R. M. & Dill, J. H. The effects of modeling cues,

reward, and attentional set on the production of grammatical and un-

grammatical syntactic constructions. Journal of ExperiMental Child

Psychology, 1968, 6, 131-140.

Zimmerman 17

Postman, L. Association and performance in the analysis of verbal learning.

In T. R. Dixon & D. L. Horton (Eds) Verbal behavior and general beha-

vior theory. Englewood Cliffs: Prentice-Hall, 1968. Pp. 550-571.

Rosenthal, R. Interprsonal expectations: effects of the experimenter's

hypotheses. In R. Rosenthal & R. L. Rosnow (Eds.) Artifact in behav-

ioral research. New York; Academic Press, 1969. Pp. 182-277.

Rosenthal, T. L., Alford, G. S., & Rasp, L. Concept attainment, generali-

zation, and xetention through observation and verbal coding. Journal

of Experimental Child Psychology, 13, in press (172).

Rosenthal, T. L., & Carroll,. . R. Factors in vicarious modification of

complex grammatical parameters. Journal of Educational Psychology,

63, in press (1972).

Rosenthal, T. L., ioore, W. B.,.Dorfman, H., & Nelson, B. Vicarious ac-

quisition of a simple concept tfith experimenter as model. Behavior

Research and Therapy, 1971, 9, 21')-227.

Rosenthal, T. L., & White, G. 1. Initial probability, rehearsal, and con-

straint in associative class selection. Journal pf.Experimental Child

Psychology, 13, in press (1972).

Rosenthal, T. L., & Whitebook, J. S. Incentives versus instructions in

transmitting grammatical parameters with experimenter .as model. Behav-

iour Research and Therapy, 1970, 8, 18)-196.

Rosenthal, T. L., & Zimmerman, B. J. Instructional E.pecfficity and outcome-

expectation in observationally-induced question formulation. Journal

of Educational Psychology, 63, in press a (1972).

2 0

Zimmerman 18

Rosenthal, T. L., & Zimmerman, B. J. npdaling by exemplification and in-

struction in training conservation. Developmental Psychology, 7, in

press b, (1972).

Rosenthal, T. L., Zimmerman, B. J., & Durning, K. Observationally-induced

changes in children's interrogative classes. Journal of Personality

and Social Psychology, 1970, 1G, 681-688.

Zimmerman, B. J. & Rosenthal, T. L. Observation, repetition, and ethnic

background in concept attainment and generalization. Child Develop-

ment, in press.

2 1

Zimmerman 19

Footnotes

1. This study was supported by the Arizona Center for Early Child-

hood Education as a Subcontractor under the National Program on Early

Childhood Education of the Central Hidwestern aegional Laboratory, a

private nonprofit corporation supported by funds from the U. S. Office

of Education, Department of Health, Education and Welfare. The opin-

ions expressed in this publication do not necessarily reflect the po-

sition or policy of the Office of Education, and no official endorse-

ment should be inferred. We wish to acknowledge the generous coopera-

tion of Principals LI. R. i4eredith, U. C. Upshaw and R. E. Whalen of

Brown, Hughes, and Whitmore Schools, of their teachers, and of the ad-

ministration of Tucson School District 1. We wish to thank John Bell,

Elaine Williams, and Joan Wyde for their assistance with aspects of

this research, and Glenn White who offered helpful comments on the

manuscript.

2. All tests of significance reported in this paper were based

on two-tailed probability estimates.

Zimmerman

Table 1

Ileans 'by Phase for Each Treatment Group and for

Combined Hodeling and Rule Variation.

20

Group

Immediate Data ! Delayed Data

Phase Phase

Base- Imita- Trans- 'mita- Trans- General-line tiou fer tion fer ization

Separate CellsI

Nodel no rule 1.63 5.42 3.92 3.81 3.67 3,92

Nodel plus rule 1.?,3 8.25 7.36 6.19 5.73 5.44

Rule no model 1.39 5.92 5.69 4.64 4.11 4,25

No model no rule _ 1.81 2.28 1.58 2.42 1.86_

1.97..

Combined Subjects

All Nodeling 1.79 6.83 5.64 5.00 4.72 4.63

All nonmodeling 1.60 4.10 3.64 3.56 2.99 3.11

All rule 1.63 7.06 6.53 5.42 4.94 4.35

All nonrule 1.75 3..05 2.75 3.14 2.76 2.94

2 3